⚡ Quick Summary
- US data centers could match New York City's daily water consumption by 2030 according to new research
- AI workloads generate far more heat than traditional computing, accelerating water demand
- Tech companies are funding hundreds of millions in local water infrastructure upgrades
- Environmental groups are calling for a moratorium on new data center construction
What Happened
A new study led by Shaolei Ren, an associate professor of electrical and computer engineering at the University of California, Riverside, has found that U.S. data centers could require enough additional water capacity by 2030 to rival New York City's entire daily water supply. The research, published on the preprint server arXiv, identifies limited public water infrastructure as an emerging critical bottleneck to data center expansion.
The findings arrive as tech giants are racing to build new data center campuses to power artificial intelligence workloads. Companies like Microsoft, Google, Amazon, and Meta have collectively committed hundreds of billions of dollars to AI infrastructure over the next several years, with much of that investment flowing into massive new facilities that require enormous amounts of water for cooling.
To address the strain on local water systems, tech companies have begun partnering with municipalities to fund water infrastructure upgrades, sometimes spending hundreds of millions of dollars. Professor Ren noted the troubling incentive dynamics at play: "Those companies are profit driven, right? So I think clearly there is something wrong," he told Gizmodo, suggesting that the public cost of AI's water appetite may be underappreciated.
Background and Context
Data centers generate tremendous heat from the continuous operation of densely packed servers, networking equipment, and storage systems. Liquid cooling is the most efficient method to prevent overheating and hardware failure, but it comes with significant water costs. Even closed-loop cooling systems that companies promote as water-efficient rely on evaporative cooling towers that consume substantial volumes of water as heat is transferred from the facility to the atmosphere.
The scale of the problem has grown dramatically with the AI boom. Training a single large language model can require millions of GPU-hours across thousands of servers, generating heat loads far exceeding traditional cloud computing workloads. Inference — running trained models to serve user queries — adds continuous thermal load that scales with adoption. As AI moves from experimental to foundational across industries, data center heat output is climbing exponentially.
Previous research by Ren's team estimated that a single ChatGPT conversation of 20 to 50 queries consumed roughly 500 milliliters of water — about the volume of a standard water bottle. Multiplied across hundreds of millions of daily AI interactions globally, the aggregate water consumption is staggering. The latest study extends this analysis to project total U.S. data center water demand through the end of the decade, factoring in announced construction plans and anticipated AI adoption curves.
The geographic distribution of data centers compounds the challenge. Many are located in regions already facing water stress — Northern Virginia, Phoenix, parts of Texas — where competition for water between data centers, agriculture, and residential users creates genuine conflict. Ensuring your business operations remain efficient regardless of infrastructure pressures starts with having the right software in place, such as an affordable Microsoft Office licence that keeps teams productive without requiring additional hardware-intensive resources.
Why This Matters
The water consumption study forces a confrontation between two powerful narratives: the transformative promise of artificial intelligence and the physical reality of the infrastructure required to deliver it. Tech companies have positioned AI as a solution to countless global challenges, including climate change and resource management. The irony that AI's own resource footprint is becoming a significant environmental concern has not been lost on regulators, environmental groups, or the public.
For local communities hosting data centers, the implications are immediate and tangible. Water rate increases, infrastructure strain, and competition with residential and agricultural users are already creating tension in data center hubs. Over 250 environmental organizations have asked Congress for a moratorium on new data center construction until water and energy impacts can be properly assessed — a request that, while unlikely to be granted, signals growing political resistance to unchecked AI infrastructure expansion.
The financial implications are significant as well. If tech companies must increasingly fund water infrastructure as a condition of data center permitting, those costs will eventually be reflected in cloud computing pricing. Enterprises relying on AI-powered services should anticipate that the resource costs of AI infrastructure may drive price increases across cloud platforms over the next several years.
Industry Impact
The data center industry is responding with a mix of technological innovation and strategic relocation. Advanced cooling technologies — including direct liquid cooling, immersion cooling, and heat reuse systems — can dramatically reduce water consumption compared to traditional evaporative approaches. Companies like Google and Microsoft have committed to water-positive operations by 2030, meaning they aim to replenish more water than their data centers consume through watershed restoration and water recycling projects.
Geographic diversification is another strategy. Data center developers are increasingly looking at cooler climates — Scandinavia, Iceland, Canada, and northern U.S. states — where ambient air cooling can reduce or eliminate water consumption entirely. However, these locations often lack the power grid capacity and fiber optic connectivity required for major data center deployments, creating a chicken-and-egg infrastructure challenge.
Nuclear power is emerging as a complementary solution. Microsoft's deal to restart Three Mile Island Unit 1 and Amazon's nuclear power investments reflect a recognition that data center growth requires clean, reliable baseload power generation that renewable sources alone cannot consistently provide. Nuclear plants can also be paired with closed-loop cooling systems that reduce freshwater demand.
For organizations running their own on-premises infrastructure, the trend reinforces the value of optimizing existing hardware and software rather than scaling up data center capacity. Running current, fully licensed software like a genuine Windows 11 key with proper security updates ensures maximum efficiency from existing equipment, reducing the need for additional hardware that ultimately contributes to data center demand.
Expert Perspective
Environmental researchers have praised the study for quantifying a problem that has been discussed qualitatively but rarely measured with this precision. The comparison to New York City's water supply — approximately 1 billion gallons per day — provides an intuitive scale reference that makes the abstract concept of data center water consumption concrete and alarming.
However, some industry analysts have noted that the study's projections assume current cooling technology and do not fully account for the rapid adoption of more water-efficient alternatives. If direct liquid cooling and immersion cooling become the default for new data center construction, actual water consumption could be significantly lower than projected.
The tension between AI progress and environmental sustainability is likely to intensify as a political and regulatory issue through 2026 and beyond. Data center permitting battles are already becoming contentious in Virginia, Georgia, and Texas, and water consumption is emerging as the most visceral point of opposition.
What This Means for Businesses
Enterprise IT leaders should expect data center sustainability to become an increasingly important factor in cloud provider selection. Companies with credible water stewardship programs and investments in advanced cooling technology will likely differentiate themselves in procurement conversations, particularly for organizations with their own ESG commitments.
The potential for cloud pricing increases driven by infrastructure costs — including water — should factor into long-term IT budgeting. Organizations that can optimize their cloud spending through efficient software licensing, workload right-sizing, and hybrid infrastructure strategies will be better positioned to absorb these cost pressures. Partnering with trusted enterprise productivity software providers ensures predictable costs for foundational business applications while cloud infrastructure pricing evolves.
Key Takeaways
- U.S. data centers could require water capacity equivalent to New York City's entire daily supply by 2030
- AI workloads generate significantly more heat than traditional cloud computing, driving water consumption higher
- Tech companies are spending hundreds of millions to upgrade local water infrastructure near data center sites
- Over 250 environmental groups have called for a moratorium on new data center construction
- Advanced cooling technologies could reduce projected water consumption but aren't yet widely deployed
- Cloud computing prices may increase as data center resource costs rise
Looking Ahead
The study's findings will likely fuel regulatory and permitting debates throughout 2026 as new data center projects enter planning and approval phases. Expect tech companies to accelerate investment in water-efficient cooling, nuclear power partnerships, and geographic diversification. The industry's ability to solve its water problem — or at least convincingly demonstrate progress — will shape public perception of AI infrastructure for years to come.
Frequently Asked Questions
Why do data centers use so much water?
Data centers use water primarily for cooling. Servers generate enormous heat from continuous operation, and evaporative cooling towers — the most common cooling method — consume large volumes of water to transfer that heat to the atmosphere.
How much water does AI use?
Research estimates a single ChatGPT conversation of 20-50 queries consumes about 500 milliliters of water. Multiplied across hundreds of millions of daily AI interactions, the aggregate consumption is enormous and growing.
Are tech companies addressing data center water usage?
Yes. Google and Microsoft have committed to water-positive operations by 2030. Companies are also investing in direct liquid cooling, immersion cooling, nuclear power, and cooler-climate data center locations to reduce water dependence.